390 John Woods
VPE based on one hundred thousand partic1es, this biological data set contains a
few trillion bytes per simulated year. In practice this is too much to diagnose, so the
VPE contains onlya small sample ofthese time series; typically one thousand phytoplankton and one thousand zooplankton particles. The user controls the criteria
used to select which partic1es will be held in the VPE. Although only a small ftaction are retained for analysis, the information carried by ali of the partic1es is used
to simulate the biological environment and biofeedback to the physical and chemical environments, and for the chronological "Parish Register" of demographic
events (18.8.1).
Two additional diagnostic products are computed after the VPE data set has been
created: demographic time series and audit trails. The demography of each population comprises time series of birth rate, death rate by each cause of death, and life
expectancy. These time series are based on the properties ofthe individuals, inc1uding the intra-population variation, which is an emergent property ofthe VPE. Such
information is not available in simulations derived by Eulerian integration.
Audit trails are time series ofthree properties of each partic1e: its depth, its ambient environment, and the biological state variables ofits sub-population. The ambient environment of a plankton sub-population is defined as the value of each
environmental variable interpolated to its exact depth (resolved to one millimetre).
Audit trails are useful for discovering what caused puzzling emergent properties of
the environment. They are not available in simulations based on Eulerian integration.
18.7.2 The Azores Virtual Plankton Ecosystem
Much of the research with VPE has been designed to simulate the plankton ecosystem at a location (41N, 27W) just north ofthe Azores. This lies on the zero line
of annual surface heat budget, i.e. where solar heating exactly balances the oceanic
cooling to the atmosphere, according to Bunker's climatology. (The corresponding
zero line for ERA c1imatology lies within lOOkm.) At such sites, the ocean environment exhibits a stationary annual cycle when driven by a stationary atmospheric
climatology. For example, every year the turboc1ine at the base of the surface turbulent (mixing) layer has the same depth on any given day ofthe year and time of
day. The annual divergence of the horizontal flux of heat is zero. The Azores also
offers other simplifying advantages: the permanent circulation is relatively slow,
the geostrophic eddy energy is low, as is annual Ekman pumping, while the
upstream gradients of the ocean environment are slack. So we can ignore the
impact of advection on the plankton ecosystem to frrst order, which simplifies analysis and verification ofVPEs created by geographically-Eulerian integration at the
Azores site. The same is not true at other mid-Atlantic sites, such as Bermuda,
where advection and eddying play important roles (McGillicuddy et al 1998).
18.7.3 Environment
The Virtual Plankton Ecosystem describes the changing environment in a water
column in terms of time series of the profiles (variation with depth) of each vari-
VPE based on one hundred thousand partic1es, this biological data set contains a
few trillion bytes per simulated year. In practice this is too much to diagnose, so the
VPE contains onlya small sample ofthese time series; typically one thousand phytoplankton and one thousand zooplankton particles. The user controls the criteria
used to select which partic1es will be held in the VPE. Although only a small ftaction are retained for analysis, the information carried by ali of the partic1es is used
to simulate the biological environment and biofeedback to the physical and chemical environments, and for the chronological "Parish Register" of demographic
events (18.8.1).
Two additional diagnostic products are computed after the VPE data set has been
created: demographic time series and audit trails. The demography of each population comprises time series of birth rate, death rate by each cause of death, and life
expectancy. These time series are based on the properties ofthe individuals, inc1uding the intra-population variation, which is an emergent property ofthe VPE. Such
information is not available in simulations derived by Eulerian integration.
Audit trails are time series ofthree properties of each partic1e: its depth, its ambient environment, and the biological state variables ofits sub-population. The ambient environment of a plankton sub-population is defined as the value of each
environmental variable interpolated to its exact depth (resolved to one millimetre).
Audit trails are useful for discovering what caused puzzling emergent properties of
the environment. They are not available in simulations based on Eulerian integration.
18.7.2 The Azores Virtual Plankton Ecosystem
Much of the research with VPE has been designed to simulate the plankton ecosystem at a location (41N, 27W) just north ofthe Azores. This lies on the zero line
of annual surface heat budget, i.e. where solar heating exactly balances the oceanic
cooling to the atmosphere, according to Bunker's climatology. (The corresponding
zero line for ERA c1imatology lies within lOOkm.) At such sites, the ocean environment exhibits a stationary annual cycle when driven by a stationary atmospheric
climatology. For example, every year the turboc1ine at the base of the surface turbulent (mixing) layer has the same depth on any given day ofthe year and time of
day. The annual divergence of the horizontal flux of heat is zero. The Azores also
offers other simplifying advantages: the permanent circulation is relatively slow,
the geostrophic eddy energy is low, as is annual Ekman pumping, while the
upstream gradients of the ocean environment are slack. So we can ignore the
impact of advection on the plankton ecosystem to frrst order, which simplifies analysis and verification ofVPEs created by geographically-Eulerian integration at the
Azores site. The same is not true at other mid-Atlantic sites, such as Bermuda,
where advection and eddying play important roles (McGillicuddy et al 1998).
18.7.3 Environment
The Virtual Plankton Ecosystem describes the changing environment in a water
column in terms of time series of the profiles (variation with depth) of each vari-